For me, I thought it was an article defending the practice of using machine learning models in unit tests